Breaks of the Game: The Incumbency Rule in an Anti-Incumbent Year

By Christopher P. Borick

One of the most interesting phenomena of preelection polls in the United States is the way undecided voters break at the end of a campaign. Since the division of undecided voters on election day can determine which candidate prevails in a close election, it is surprising how little research has directly focused on this important question.

Perhaps the most widely accepted, if not empirically proven, theory regarding the impact of incumbency is that undecided voters behave differently when an incumbent is in a race. When an incumbent seeks reelection, voters are, in essence, asked to evaluate his or her record in office. As Guy Molyneux argues,

Elections are fundamentally a referendum on the incumbent. The first step in voters’ decision-making process is to answer the question “does he deserve re-election?” Undecided voters have basically answered that question in the negative, and their undecided status reflects the fact that they don’t know enough about the challenger (yet) to feel comfortable stating a public preference.

Theoretically, then, the incumbent is at a substantial disadvantage in securing undecided votes. Voters who are conflicted in the last days and hours before an election will be particularly hard for an incumbent to sway; thus, the “incumbent rule” holds that undecided voters will break for the challenger as the race draws to a close.

Followers of preelection polls often use the “incumbent rule” to help predict election outcomes. Press coverage regularly assumes that an incumbent with less than 50 percent support in the polls will be in jeopardy on election day. As undecided voters break to the challenger, incumbents will receive a portion of the vote that is at or below the level they received in the final election polls. This phenomenon seems to have been fairly well supported in recent presidential elections (see Table 1).     

While incumbents in presidential races do track close to their preelection poll numbers, we should be careful not to generalize to other races. Since the small number of cases and relatively high levels of turnout and public attention make presidential races exceptional in nature, it’s important to look at a wider range of elections to see how robust the incumbent rule may be. During the past two decades a number of studies have examined the way undecided voters break at the end of an election and the role that incumbency plays in these decisions.

The first study to examine the incumbent rule empirically was undertaken by Nick Panagakis in 1989. Panagakis examined 155 races, primarily from the mid- to late 1980s, to determine how undecided voters broke at the end of elections when an incumbent was present in the campaign. He found that in 127 out of the 155 cases in his sample, the challenger won a majority of the undecided voters. Conversely, there were only 19 races where the incumbent took most of the undecided voters (Table 2).   

Panagakis offered a number of explanations for his findings. He contended that

undecided voters are not literally undecided, not straddling the fence unable to make a choice—the traditional interpretation. An early decision to vote for the incumbent is easier because voters know incumbents best. It helps to think of undecided voters as undecided about the incumbent, as voters who question the incumbent’s performance in office.

In the Panagakis study the challengers who did not get a majority of the undecideds appear to be candidates who had held office of equal stature to the one they were seeking. Because voters were as familiar with the challengers as the incumbents, the challengers assumed the same characteristics as the incumbents, negating the advantages they otherwise would have had.

While Panagakis found a fairly robust advantage for challengers in the division of undecided voters in 1989, a study by Chris Bowers in 2004 shows a trend of increasing gains for incumbents.

As Figure 1 shows, between 1976 and 1988, undecided voters broke by a margin of four to one. Between 2000 and 2004, this margin declined to about three to two. Bowers attributes the narrowing gap to increasingly ideological political parties. He argues that the difference in parties is “becoming starker,” and, as a result, “more people have made up their minds going into the booth.”

Despite this apparent change, both the Panagakis and Bowers studies demonstrate an overall tendency of challengers to gain most of the undecided voters in the final weeks of an election. While providing valuable insight into the general nature of undecided voter allocation, however, the Panagakis and Bowers studies do not pay substantial attention to the affects of factors such as partisanship on undecided voter decisions. In this study, I expand upon the earlier ones by both updating the findings to include the 2006 elections and by examining the effect of partisanship on the “incumbency rule.”

The first step is to examine the elections in 2004 and 2006 by the same standards used by Panagakis and Bowers in their earlier studies. Both Panagakis and Bowers tallied which candidate (incumbent or challenger) won the greater share of the undecided votes during the elections they examined.

Drawing poll results from Real Clear Politics and The Polling Report, I began by subtracting the final averaged poll results from senate and gubernatorial polls for the incumbent from the actual election result for the incumbent to determine the total number of points the incumbents gained from the final poll until the election. Polls included in this measure had at least one day of the polling period occur during the seven days before the election. I performed the same calculations for the challengers

Next, I added the totals from these two steps and divided the result back into the incumbent and challenger totals in order to calculate the percentage of unallocated voters for each category.

As can be seen in Figure 2, the 2004 results conform very closely to the averages Bowers found for races between 2000 and 2004. More specifically, the incumbent candidate took 41 percent of undecided votes in this study, compared to 40 percent in 2000 and 41 percent in the combined 2002 and 2004 elections. In 2006, however, there was a substantial increase in the performance of incumbents with regard to undecideds, with candidates in office taking almost half of unallocated voters from the preelection polls. This 49 percent incumbent result was the highest found in any of the studies, and marked a noticeable shift from the earlier election periods studied by Panagakis and Bowers.

While the results in Figure 2 do demonstrate more success for incumbents in winning late undecided voters, they do not capture the effect of the candidates’ party affiliation. The 2006 midterms were notable for the fact that not a single Democratic incumbent lost a race at the gubernatorial, senatorial, or congressional level. With high presidential disapproval rates placing a drag on GOP candidates, the 2006 campaign became more secure for Democratic incumbents. Of course, it was also a difficult year for Republican incumbents, with the Democrats winning control of both the House and Senate.

Although the effect of partisanship on the overall election outcome is obvious, the effect of party affiliation on undecided voters is less clear. The question in particular is, will a partisan “wave” wash away the general advantages that challengers have in winning undecided voters? In the case of 2006, the answer appears to be yes. As can be seen in Figure 3, both Democratic challengers and incumbents won a majority of undecided voters. This scenario contrasts with 2004, when both Democratic and Republican challengers took a majority of undecided voters.

Another way to examine the allocation of undecided voters is to measure the size of the deviation between preelection polls and the actual election results. In the 2004 general election, this measure followed a pattern that would be expected given previous research on the matter. More specifically, incumbent poll predictions deviated by about three percentage points from actual election results, while challenger predictions were about four and a half points off the actual results. These margins form a ratio of about three to two in terms of undecided voters breaking to challengers. While this margin is much less than is commonly assumed in media accounts of how undecided voters break, the results are in line with what Bowers found in his examination of elections between 2000 and 2004.

Looking more closely at the data in Table 3, it appears that candidate and election characteristics were not important factors in determining the deviation of poll predictions from actual results. Democratic and Republican incumbents had nearly identical levels of deviation—3.08 and 3.10 points, respectively—although Democrats were successful in gaining a larger share of unallocated voters than Republicans. The type of race seemed to have limited effect on the share of undecided voters won by candidates, with only Democratic senatorial incumbents outperforming the overall incumbent average.

While in 2004 undecided voters were more likely to break to challengers over incumbents, in 2006 the split was considerably more even. For incumbents, preelection polls deviated by 2.18 points from actual results, while challengers deviated by 2.34 points. This limited difference underlies the diminished role that incumbency status appears to have played in the 2006 midterms.

However, the aggregate measures mask the interactive effect of incumbency and partisanship. As can be seen in Table 4, Democratic incumbents and challengers outperformed their Republican counterparts by 8 and 14 percent, respectively. Put in another way, preelection polls were more accurate at predicting the final results of candidates challenging Democratic incumbents (1.92) than they were the results of Democratic incumbents themselves (2.29). Such evidence supports the argument that the incumbent rule broke down under the Democratic wave in 2006.

The allocation of undecided voters plays a central role in analyses of preelection polls. Reporters regularly query pollsters on how undecided voters break during the final weeks and days of a campaign and look for direction on the impact of undecideds on election outcomes. To address these questions some common assumptions appear regularly in media accounts about polls. In particular, great weight is often placed on the percentage of the vote that an incumbent receives in the polls because it is assumed that undecided voters will break overwhelmingly to the challenger. Thus, no matter how large the spread between incumbent and challenger, incumbents polling below 50 percent can be considered in jeopardy of losing their seats. Studies of presidential elections support this contention, with incumbents’ poll numbers deviating by no more than two percentage points between 1980 and 2000.

However, in nonpresidential elections, the share of undecided voters that appears to go to incumbents is significantly greater than in presidential races, and, according to Bowers, this percentage appears to be rising over time. The evidence presented here supports the continuing decline of “the incumbent rule” regarding undecided voters. In the aggregate elections of 2004–6, challengers received slightly over half of unallocated voters in preelection polls.

Even more pronounced was the near even split of undecided voters between incumbents and challengers during the 2006 midterm elections. While incumbents themselves garnered about the same share of unallocated voters in 2006 as challengers, the interactive effect of partisanship and incumbency in this race appears to have been even more notable. More specifically, the aggregate incumbency-challenger split masked significant differences in how partisanship determined the break of undecided voters in 2006. This study finds that while Republican incumbents took a share (43 percent) of unallocated voters that was consistent with overall incumbent performances since 2000, Democratic incumbents actually took a majority (54 percent) of the undecided voters in 2006.

This study provides continuing evidence of the weakening of the “incumbent rule.” In nonpresidential elections, it is no longer viable to make claims that undecided voters will break solidly towards challengers, and that incumbents’ final results will match their preelection poll numbers. References made by both the media and pollsters to the incumbent rule are citing a phenomenon that no longer exists.

It may now be more valuable to complete a more comprehensive empirical analysis of the preelection polls to see if the break of unallocated voters may be tied to some “bandwagon” or “wave” effect that overrides any perceived challenger advantage. We might ask, for example, whether in years like 2006 and 1994 there was a tipping point where the overall strength of one party pushed undecided voters at a level equal to or greater than their knowledge and familiarity with incumbent candidates. To answer this question more empirical analysis will have to be undertaken. In the meantime, the “incumbent rule” deserves to go on hiatus.

Christopher P. Borick is director of the Muhlenberg College Institute of Public Opinion.

Polls in the Panagakis Study

The Pangagakis study collected and analyzed polls that were mostly conducted within two weeks of election day. They covered both general and primary elections with Democratic and Republican incumbents and were predominantly from statewide races, with a few U.S. House, mayoral, and countywide contests included. Most dated from the 1986 and 1988 elections, although a few came from the 1970s. The polls included in the study were drawn from CBS, Gallup, Gordon S. Black Corp., Market Opinion Research, Tarrance Associates, and Mason-Dixon Opinion Research, as well as polls that appeared in The Polling Report.

Polls in the Bowers Study

The Bowers study collected and analyzed polls in which at least one day of the polling period occurred within the last week before election day. To determine the allocation of undecided voters, Bowers employed a three-step process. First, he subtracted the final poll result for the incumbent from the actual election result for the incumbent to determine the total number of points the incumbent gained from the final poll until the election. He then added up all 451 of these results to determine the total number of points all incumbents gained from all final polls until the election. In his second step he repeated the process for challengers. In step three he took the total from step one and added it to the total from step two. He then divided this total into both the results from step one and step two to determine the relative gain for the incumbent and the challenger.

Additional Reading

Blumenthal, Mark. 2004. Incumbent rule redux. Mystery Pollster. August 14.

Bowers, Chris. 2004. Incumbent rule research update. MyDD Direct Democracy. September 3.

Fenwick, Ian, Frederick Wiseman, John F. Becker, and James R. Heiman. 1982. Classifying undecided voters in pre-election polls. Public Opinion Quarterly 46.

Lavarkas, Paul, and Michael W. Traugott. 2000. Election polls, the new media and democracy New York: Chatham House.

Mitofsky, Warren J. 1999. Reply to Panagakis. Public Opinion Quarterly 63, no. 2 (Summer, 1999): 282-84

Molyneux, Guy. 2004. The big five-oh. American Prospect: Online Edition. October 1.

 
Panagakis, Nick. 1989. Incumbent races: Closer than they appear. Polling Report. February 27.